How much does screen size matter in comparing Ryzen Mobile and Kaby Lake-R battery life?


— 1:15 PM on November 30, 2017

 As we've continued testing AMD's Ryzen 5 2500U APU over the past few days, we've been confronted with the problem of comparing battery life across laptops with different screen sizes. Many readers suggested that I should take each machine's internal display out of the picture by hooking them up to external monitors. While I wanted to get real-world battery-life testing out of the way first, I can certainly appreciate the elegance of leveling the playing field that way. Now we have.

Before we get too deeply into these results, I want to point out loudly and clearly that these numbers are not and will never be representative of real-world performance. Laptop users will nearly always be running the internal displays of their systems when they're on battery, and removing that major source of power draw from a mobile computer is an entirely synthetic and artificial way to run a battery life test. We're also still testing two different vendor implementations of different SoCs, and it's possible that Acer's engineers might have some kind of magic that HP's don't (or vice versa). Still, for folks curious about platform performance and efficiency, rather than the more real-world system performance tests we would typically conduct, these results might prove interesting.

To give this approach a try, I connected both the Envy x360 and the MX150-powered Acer Swift 3 to 2560x1440 external monitors running at 60 Hz using each machine's HDMI output. I then configured each system to show a display output on the external monitor only and confirmed that both laptops' internal displays were 100% off. After those preparations, I ran our TR Browserbench web-browsing test until each machine automatically shut off at 5% battery before recording their run times.

As we'd expect, both machines' battery life benefits from not having to power an internal monitor. Counter to our expectations, though, the Envy x360 doesn't actually seem to spend a great deal of its power budget on running its screen. The Envy gained only 53 minutes, or 15%, more web-browsing time than when it didn't have to drive its internal monitor. The MX150-powered Acer, on the other hand, gained a whopping five hours of battery life when we removed its screen from the picture. I was so astounded by that result that I retested the Envy to ensure that a background process or other anomaly wasn't affecting battery life, but the HP machine repeated its first performance.

We can take battery capacity out of the efficiency picture for this light workload by dividing minutes of run time by the capacity of the battery in watt-hours. This approach gives us a normalized "minutes per watt-hour" figure that should be comparable across our two test systems. HWiNFO64 reports that the Envy x360 has a 54.8 Wh battery, and since it's brand-new, a full charge tops up that battery completely. Using the technique described above, we get 7.8 minutes of run time per watt hour from the HP system.

The Acer Swift 3 I got from Intel appears to have been a test mule at some point in its life. HWiNFO64 reports that the Swift 3 has already lost 10% of its battery capacity, from 50.7 Wh when it was new to 45.7 Wh now. In this measure of efficiency, though, that capacity decrease actually helps the Swift 3. The system posts a jaw-dropping 19 minutes of run time per watt-hour for light web browsing, or a 2.4-times-better result.

Although this is a staggering difference, I emphasize that it's not representative of performance in the real world. If we don't remove the display from the picture, the Optimus-equipped Swift 3 only posted nine and a half hours of run time in our i5-8250U review, or only about half again as long as the Envy's six hours and 12 minutes. If we drop the MX150 from the picture, the IGP-only Swift 3s and their 10.5 hours of battery only run 67% longer than the Envy. Those are only rough assessments of platform potential, given that we aren't normalizing for battery capacity or screen size. Still, Ryzen Mobile systems might have a ways to go to catch Intel in the battery life race. The blue team has been obsessed with mobile power management for years, and technologies like Speed Shift are just the latest and most visible results of those efforts.

In any case, it's clear that there's a lot of moving parts behind the battery life of these systems. I've repeatedly cautioned that it's early days for both drivers and firmware for the Ryzen 5 2500U, and it's possible that future refinements will close this gap somewhat. Benchmarking a similar Intel-powered system from HP might also help even the field, given my research in my first examination of the Ryzen-powered Envy x360's battery life. (If you'd like to help with that project, throw us a few bucks, eh?) Still, if you favor battery-sipping longevity over convertible versatility and raw performance, it seems like the Envy x360 requires a compromise that our GeForce-powered Acer Swift 3 doesn't. Stay tuned for more battery-life testing soon.

Tip: You can use the A/Z keys to walk threads.
View options